Estimating bamboo forest aboveground biomass using EnKF-assimilated MODIS LAI spatiotemporal data and machine learning algorithms

High-precision LAI (leaf area index) spatiotemporal data obtained from MODIS satellite remote sensing products are important for studying vegetation growth status, biomass carbon reserves, and the spatiotemporal dynamics of carbon cycling. LAI significantly influences biomass accumulation during the growth of bamboo forest in subtropical zones. Therefore, we applied the ensemble Kalman filter (EnKF) data assimilation algorithm to assimilate MODIS LAI products, and used assimilated LAI and the normalized difference vegetation index, enhanced vegetation index, simple ratio index as variables in the random forest model to estimate bamboo forest above ground biomass (AGB) in Zhejiang Province. Assimilated LAI spatiotemporal data using EnKF greatly improve the accuracy of MODIS LAI products, the R2 between assimilated and observed LAI was 0.92, and the RMSE was 0.37. Variations in the assimilated LAI time series were consistent with the seasonal dynamics of bamboo forest growth and had a significant effect on AGB. Moreover, the random forest model had strong predictive capabilities. A comparison of training and testing results produced accuracy (R) values for the random forest model using the assimilated LAI time series of 0.71 and 0.73, respectively. Using the assimilated LAI achieved a more accurate AGB estimate than using MODIS LAI time series products, as the R values were 54.3% and 58.7% higher, and the RMSE values were 19.2% and 19.1% lower for training and testing results, respectively. The calculated spatial distribution of bamboo forest AGB in Zhejiang province was consistent with the observed values. By combining assimilation technology of the MODIS LAI time series with the random forest model to more accurately estimate bamboo forest AGB in Zhejiang province, this study provided a new method for estimating large scale forest AGB based on low-resolution time series data.